Association between body adiposity index and coronary risk in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil)

Association between body adiposity index and coronary risk in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil)

Clinical Nutrition xxx (xxxx) xxx Contents lists available at ScienceDirect Clinical Nutrition journal homepage: http://www.elsevier.com/locate/clnu...

893KB Sizes 0 Downloads 36 Views

Clinical Nutrition xxx (xxxx) xxx

Contents lists available at ScienceDirect

Clinical Nutrition journal homepage: http://www.elsevier.com/locate/clnu

Original article

Association between body adiposity index and coronary risk in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil) rio Tosta de Almeida a, b, *, Alexandre da Costa Pereira c, Roge Maria de Jesus Mendes da Fonseca d, Sheila Maria Alvim de Matos a, ~o Aquino a Estela Motta Lea a

Instituto de Saúde Coletiva, Universidade Federal da Bahia, Salvador, BA, Brazil Departamento de Saúde, Universidade Estadual de Feira de Santana, Feira de Santana, BA, Brazil ~o, Hospital das Clínicas/FMUSP, Sa ~o Paulo, SP, Brazil Instituto do Coraça d Escola Nacional de Saúde Pública/FIOCRUZ, Rio de Janeiro, RJ, Brazil b c

a r t i c l e i n f o

s u m m a r y

Article history: Received 15 September 2018 Accepted 3 June 2019

Background & aims: The body adiposity index (BAI) was recently proposed as a better indicator of body adiposity than body mass index in adults. The association between BAI and cardiometabolic risk factors has been widely investigated. However, the strength and magnitude of these associations varied as a function of the endpoint evaluated, the study design, the population investigated, and the cut-off points used. The aim of this study was to investigate the association between BAI and coronary heart disease (CHD) risk in a large sample of Brazilian adults and to propose the most appropriate cut-off points for BAI for the identification of CHD risk in the adult Brazilian population. Methods: Data from 15,092 civil servants (54.4% women) from universities and research institutes in six Brazilian states were evaluated in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil). All were aged 35e74 years at baseline. The Framingham coronary risk score was used to identify a very high risk of CHD (20% risk of CHD over the next 10 years) and a high risk of CHD (10% risk). BAI cut-off points capable of detecting a risk of CHD were determined using ROC curves and associations were tested using Poisson regression with robust variance, according to sex and age. Three multivariable models were examined. Results: BAI remained positively associated with a very high and high risk of CHD following adjustment for potential confounding factors in all the strata and multivariable models (p < 0.05), with the exception of model 3 (adjusted for education level and waist-to-hip ratio) for very a high risk of CHD in younger women (p ¼ 0.06). In the adjusted models, the prevalence ratios for a very high and high risk of CHD, irrespective of age group, varied between 1.23 (95%CI: 1.09e1.39) and 1.64 (1.33e2.03) and 1.07 (1.03 e1.12) and 1.47 (1.36e1.60) in men; and 1.57 (1.08e2.31) and 2.42 (1.36e4.31) and 1.29 (1.13e1.47) and 1.82 (1.54e2.15) in women, respectively. The optimal cut-off points of BAI to determine a risk of CHD were: 28.0 in men of both age groups, and 34.0 in younger women and 36.0 in older women to determine a very high risk; and 26.0 in younger men and 34.0 in women of both age groups to determine a high risk. BAI showed a reasonable ability to predict coronary risk in participants of the ELSA-Brasil (AUC>60%, except for the group of men of 60e74 years of age). Conclusions: Higher BAI levels were found to be associated with a greater risk of developing CHD in both men and women of different ages participating in the ELSA-Brasil, suggesting that BAI may be a useful tool for screening for CHD risk in Brazilian adults. © 2019 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.

Keywords: Body adiposity index Obesity Anthropometry Risk factors Epidemiological study

Abbreviations: BAI, body adiposity index; CHD, coronary heart disease; CVD, cardiovascular disease; BMI, body mass index; T2DM, type 2 diabetes mellitus; FRS, Framingham risk score; WHR, waist-to-hip ratio; IPAQ, International Physical Activity Questionnaire; PR, prevalence ratio; AIC, Akaike Information Criterion; AUC, area under the ROC curve; SBP, systolic blood pressure; DBP, diastolic blood pressure. * Corresponding author. Instituto de Saúde Coletiva, Universidade Federal da Bahia, Av. Araújo Pinho, 513, Canela, 40110-150, Salvador, BA, Brazil. E-mail addresses: [email protected] (R.T. Almeida), [email protected] (A.C. Pereira), [email protected] (M.J.M. Fonseca), sheilaalvim@ gmail.com (S.M.A. Matos), [email protected] (E.M.L. Aquino). https://doi.org/10.1016/j.clnu.2019.06.001 0261-5614/© 2019 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.

Please cite this article as: Almeida RT et al., Association between body adiposity index and coronary risk in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil), Clinical Nutrition, https://doi.org/10.1016/j.clnu.2019.06.001

2

R.T. Almeida et al. / Clinical Nutrition xxx (xxxx) xxx

1. Introduction

2. Methods

Body adiposity index (BAI) was proposed in 2011 as a better indicator of body adiposity than body mass index (BMI) in adult Americans [1]. Validation studies [2e4] involving comparison with other more precise methods of measuring adiposity indicate that the accuracy of BAI is similar to that of other indicators, differing as a function of the population evaluated. BAI has been found to perform better than BMI in some population groups [3,5e9]; however, not in others [2,4,10e12]. Although Bergman et al. [1] did not investigate the use of BAI as a predictor of health risks and outcomes, later studies tested the association between BAI and arterial hypertension [13e17], type 2 diabetes mellitus (T2DM) [18e21], insulin resistance [22], metabolic syndrome [23e25], cardiovascular disease (CVD) risk [26], a set of two or more cardiometabolic risk factors [2,27] and mortality [28,29]. The strength and magnitude of these associations varied as a function of the outcome, the study design and the population investigated, as well as the cut-off points used [2,13e29]. Global coronary risk assessments such as the Framingham risk score (FRS) incorporate risk markers, generating a score that allows the interaction between these factors to be calculated, enabling an individual's risk of developing coronary heart disease (CHD) to be predicted. However, few studies have analyzed the association between BAI and the FRS [30,31]. Another question that requires better answers refers to how the anthropometric indicators of general and central obesity are associated with coronary risk. The measurement of hip circumference alone has been negatively associated with cardiovascular disease in some prospective studies; however, the protective metabolic effects associated with greater hip circumference are boosted when hip circumference is adjusted for other anthropometric indicators such as BMI or waist circumference [32]. Therefore, the optimal relationship between hip circumference and other anthropometric measurements for establishing health risks remains unclear. Nevertheless, anthropometric indicators that incorporate hip circumference in their calculation appear to be potentially good predictors of health risk in large populations. Waist-to-hip ratio (WHR), a recognized indicator of central obesity, has been associated with cardiometabolic risk factors in some populations [33,34]. Establishing whether BAI, an indicator of general obesity, is associated with coronary risk irrespective of the effect of the WHR is of great relevance. However, to the best of our knowledge, no studies have assessed this question. The results of new studies conducted with different populations are essential to enable comparisons to be made that would broaden this debate. Studies conducted in Brazil on the association between BAI and cardiometabolic risk factors involved specific groups [24,26] or failed to investigate global coronary risk [15,17,20,24,35]. The usefulness of applying BAI in clinical practice and in community diagnosis of coronary risk should be established to enhance the effectiveness of actions aimed at improving the level of health in the adult Brazilian population. Hence, the objectives of the present study were: a) to investigate the association between BAI and CHD in a large sample of Brazilian adults; b) to test whether there is an association between BAI and CHD irrespective of central obesity assessed according to WHR; and c) to propose the most appropriate cut-off points for BAI for the identification of CHD in the adult Brazilian population.

2.1. Study population The Brazilian Longitudinal Study of Adult Health (ELSA-Brasil) is a prospective cohort study conducted with civil servants from teaching and research institutes in six Brazilian cities (Belo Hori ria). zonte, Porto Alegre, Rio de Janeiro, Salvador, S~ ao Paulo and Vito The objective is to investigate the incidence and progression of cardiovascular disease and diabetes and the biological, behavioral, occupational, psychological, environmental and social determinants of these diseases. At baseline (2008e2010), all active or retired employees of 35e74 years of age at the six institutions were eligible for inclusion in the study, reaching a total of 52,137 potential participants in 2008. Sample size estimation was based on the main study outcomes: T2DM and myocardial infarction (cardiovascular disease). Considering an alpha value of 5%, statistical power of 80%, exposure prevalence of 20%, and a relative risk of 2.0, the necessary sample size was estimated at approximately 6400 subjects. To enable gender-specific analyses to be performed and to allow for possible losses to follow-up, the desired sample size was defined at approximately 15,000 individuals. A total of 15,105 participants were enrolled. The methodology used in the study has already been described elsewhere [36]. To calculate the cut-off points, the present study included the data of 15,092 participants. Individuals for whom the data required to calculate the BAI and WHR were incomplete were excluded from the study (n ¼ 9), as were those using silicone implants for gluteal augmentation (n ¼ 4). A random subset of the cohort, including approximately 10% of the participants (1,541 subjects), was used to validate cut-off points, taking the distribution across the six study centers as well as biological and socioeconomic characteristics of the participants into consideration. In addition, participants with no data available on the other covariables included in the analyses of association: ethnicity/skin color (n ¼ 182), leisure time physical activity (n ¼ 216), smoking (n ¼ 1) and dietary patterns (n ¼ 20) were excluded from the analysis (n ¼ 419) (Supplementary Table 1). All the other analyses were performed with 14,673 participants (97.1% of the sample). This study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving human subjects were approved by the internal review boards of all six participating institutions and by the National Committee of Ethics in Research of the Brazilian National Health Council (CONEP). Written informed consent was obtained from all participants. 2.2. Production of demographic, anthropometric and clinical data Data collection involved interviews and measurements, all performed by a duly trained team under the supervision of qualified professionals. Procedure manuals were developed to optimize the standardization of these measurements. Standardized questionnaires were used to obtain data on demographic characteristics such as age, sex, ethnicity/skin color and education level, as well as smoking, leisure time physical activity and dietary patterns. Physical activity was measured using the long form of the International Physical Activity Questionnaire (IPAQ), taking current habits reported at the time of the interview into consideration. In the present study, only the “leisure time” domain was analyzed. Individuals who reported performing more than 150 min of walking/moderate intensity physical activity per week and/or more than 60 min of vigorous intensity activity per week and/or more

Please cite this article as: Almeida RT et al., Association between body adiposity index and coronary risk in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil), Clinical Nutrition, https://doi.org/10.1016/j.clnu.2019.06.001

R.T. Almeida et al. / Clinical Nutrition xxx (xxxx) xxx

than 150 min of any combination of walking, moderate or vigorous intensity activity per week were classified as “physically active”. Dietary patterns were defined based on data from the previously validated ELSA-Brasil food frequency questionnaire, which includes a total of 114 food items. The number of patterns was selected in accordance with the best separation of the subgroups of food items in the cluster graph. These subgroups were previously created by multiple correspondence analysis. Four dietary patterns were identified: 1) “bakery products” (consisting of the daily consumption of refined grains, bread, cookies, fried chicken, whole milk and dairy products, with no consumption of vegetables, fruits or legumes); 2) “low sugar/low fat” (characterized by the daily or weekly consumption of whole grains, skimmed milk and unsweetened drinks or soya-based drinks); 3) “fruit and vegetables” (consisting of the daily consumption of raw fruits and vegetables, fruit, grilled chicken, white cheese and skimmed milk, with no consumption of red meat, beans, refined grains or confectionary in general); and 4) “traditional” (including daily consumption of refined grains and cereals such as rice, for example, and no consumption of fast foods). Further information on the methodology used to define dietary patterns is available in a previously published paper [37]. A standardized protocol was developed for the study based on the recommendations of the International Society for the Advancement of Kinanthropometry (ISAK) to serve as guidance for the anthropometric measurements. Height, waist circumference and hip circumference were measured following overnight fasting (8e12 h), with the participants in an upright position, barefoot and wearing standardized clothing supplied by the ELSA-Brasil. Height was measured to the nearest 0.1 cm using a stadiometer (Seca-SE-216, Seca Brasil, Brazil). Waist and hip circumference were measured using a 200-cm non-stretchable tape measure (CESCORF). Measurements were taken in triplicate and the mean of the three measurements was used in the analysis. Hip circumference was measured to the nearest 0.1 cm, at the maximum protrusion of the gluteal muscles (hip), while waist circumference was measured at the midpoint between the lower border of the ribs and the iliac crest along the midaxillary line. Blood samples were taken to measure glucose, total cholesterol and its fractions following 12-h overnight fasting. Blood sampling was performed in the individual research centers and the material was stored in dry tubes. To guarantee the quality and standardization of the results, all the samples were processed and analyzed at a central laboratory established for the ELSA-Brasil. Arterial pressure was measured three times following five minutes of resting, with intervals of one minute between each measurement. Participants were fasting and with an empty bladder, according to the standardized protocol developed for the study. Measurement was performed using an automated blood pressure monitor (OMRON, model HEM-705 CP). 2.3. Definition of the anthropometric indicators and evaluation of coronary risk BAI was first proposed in 2011 as a better indicator of body fat than BMI in black and Mexican-American adults (men and women) of 18e67 years of age [1]. Calculation is based on the following equation: BAI ¼ (hip circumference in centimeters/(height in meters)1.5)-18). WHR was obtained by dividing waist circumference by hip circumference in cm. Central obesity was defined as WHR 0.90 in men and 0.85 in women [38]. Coronary risk was identified from the FRS according to the proposals published by Wilson et al. [39], using a Cox regression model that included age, systolic blood pressure (SBP), diastolic blood pressure (DBP), total cholesterol,

3

high-density lipoprotein cholesterol (HDL-C), smoking and diabetes. Participants who at the interview reported being current cigarette smokers were defined as smokers. The presence of diabetes was defined as: a previous medical diagnosis of the disease, fasting glucose 126 mg/dl or impaired glucose tolerance test with levels 200 mg/dl after 2 h of glucose load, or glycated hemoglobin 6.5%. Coronary risk was classified into two different levels. First, if the FRS was indicative of a 20% or greater 10-year risk of developing CHD, i.e. a score of 9 for men and a score of 15 for women, this was classified as a “very high risk of CHD”. If the FRS was above 6 for women and above 10 for men, representing a 10-year risk of coronary disease of 10% or more, this was classified as a “high risk of CHD”. To classify the other risk factors, the criteria proposed by Wilson et al. [39] to define very high risk were used (elevated total cholesterol:  280 mg/dl, elevated blood pressure: SBP  160 mmHg and/or DBP  100 mmHg, and low levels of HDLC: <35 mg/dl). 2.4. Statistical analysis A descriptive analysis was performed, including measures of central tendency and dispersion. The normality of the data was tested descriptively and by graphic analysis (histogram and P-plot). The ShapiroeWilk statistical test was applied to clarify any doubts. Variance was tested using the F test for homogeneity of variances. Statistical significance was defined as p < 0.05. The optimal cut-off points for the BAI were calculated for the detection of a very high or high risk of CHD using receiver operating characteristic (ROC) curves, according to sex and age group. The Youden index (J), which assesses sensitivity and specificity, was used to select the optimal cut-off point. The validity of these points was tested in a subset of the cohort. Men and women were analyzed as separate groups right from the onset on the grounds of biological interaction. The body fat component of women is higher than that of men and, furthermore, the Framingham score takes the effect of sex into consideration in the identification of CHD. Analysis of the effect modification was based on the stratumspecific measures and their respective 95% confidence intervals. The ManteleHaenszel test was used to assess the homogeneity of the prevalence ratios (PR) between the strata for each variable with a 5% significance level. Clear differences in risk ratio were found with respect to age in men (p ¼ 0.002). In women, this initial evidence was not found (p ¼ 0.327). Nevertheless, in later analyses, the crude and adjusted measures of effect showed that age considerably modified the risk ratio, strengthening the idea of an interaction with the endpoint, previously indicated by sex. Therefore, the groups were stratified by age and sex. Analysis of the potential confounding factors (education level [0 ¼ university degree; 1 ¼ elementary or high school or incomplete university education]; ethnicity/skin color [0 ¼ other; 1 ¼ brown/black-skinned]; leisure time physical activity [0 ¼ insufficiently active; 1 ¼ physically active] and dietary pattern [0 ¼ bakery; 1 ¼ low sugar/low fat; 2 ¼ fruits and vegetables; and 3 ¼ traditional]) was made firstly by observing their behavior in the associations with the endpoint and principal predictor in each stratum according to sex (male or female) and age group (<60 years or  60 years). The modeling process was performed using the backwards procedure, with possible confounding factors associated at p  0.10 remaining in the model. Other smaller models were examined, with the variables believed to be confounding factors of the principal association being removed one by one. The fitness of the models was tested using Pearson residuals, with models with

Please cite this article as: Almeida RT et al., Association between body adiposity index and coronary risk in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil), Clinical Nutrition, https://doi.org/10.1016/j.clnu.2019.06.001

4

R.T. Almeida et al. / Clinical Nutrition xxx (xxxx) xxx

p < 0.05 being considered inappropriate. The multiple models with lower Akaike Information Criterion (AIC) values for each stratum were believed to provide the best explanation for the association between BAI and a 20% or greater risk of coronary disease, with the exception of the group of older women, since, in the model recommended for this criterion, all the adjustment variables (education level, ethnicity/skin color and leisure time physical activity) had a p-value >0.05. Results of the modeling procedure are shown in Supplementary Table 2. The associations between BAI and a 10% or greater risk of CHD were tested using the same predictive models. The simple and multiple associations between coronary risk (the endpoint) and the principal exposure variable (BAI) were tested using Poisson regression with a robust error variance. Four models were examined: a crude model (unadjusted); multivariable model 1 (adjusted for education level, leisure time physical activity and dietary pattern in men; education level in younger women; education level and leisure time physical activity in older women); multivariable model 2 (adjusted only for WHR); and multivariable model 3 (adjusted for variables in model 1 plus WHR). All the statistical analyses were performed using the Stata software program, version 12.0, with statistical significance set at p < 0.05. 3. Results The cut-off points were proposed based on an analysis of 6,881 men and 8,211 women. In the other analyses, 6,681 men (22.3% of 60 years of age or more) and 7,992 women (21.1% of sixty years of age or more) were analyzed. Overall, the frequency of risk factors for CHD (high and very high risk of CHD, elevated blood pressure, lower levels of HDL-cholesterol, and diabetes) was greater in men compared to the women in the same age group. In the older age group, however, elevated total cholesterol levels were more common in women than in men, although the mean age of the men in this age group was slightly higher than that of the women. For both men and women, not only mean age but also BAI, a high and a very high risk of CHD, total cholesterol, and blood pressure were higher in the participants of 60e74 years of age compared to the younger participants, with the exception of total cholesterol in men. Coronary risk and diabetes were more common in participants of 60e74 years of age and in men, with a greater percentage of older men having a 20% or greater or 10% or greater risk of CHD and diabetes compared to the other groups. Smoking was more common among the younger participants. The frequency of young men with poorer schooling (only elementary or high school) was greater than that of the women in the same age group; however, in the group of older participants the women were more likely to have poorer education levels. Women in both age groups and men in the 35-59-year age group were more likely to report insufficient leisure time physical activity. Dietary patterns differed between men and women and between age groups (Table 1). The optimal cut-off points of BAI to determine a 20% or greater risk of CHD in men, identified from the highest Youden's index value, was the same for the entire group of men and for the group of older men for the determination of a 10% or greater risk of CHD, with 28.0 being the score suggested. For younger men, a BAI score of 26.0 best identified a high risk of CHD. A BAI score of 34.0 was selected as the optimal cut-off point with which to identify a high and a very high risk of CHD for the entire group of women and for the group of younger women, with BAI 36.0 being selected to identify a very high risk of CHD in the group of older women. The validity of the proposed cut-off points was tested in a random sample from the cohort. It was only in the group

of older men that the adequacy of the optimal cut-off point (AUC: 0.594; 95% CI: 0.483e0.705) was not confirmed (Table 2). Evaluating and comparing the areas under the ROC curve (AUC) between sex and age group, the discriminatory power of the BAI for identification of a very high risk of CHD was better for the women than for the men in 60-74-year age groups, and for the younger men as compared to the older individuals (Fig. 1). Cut-off points for other risk factors of CHD, with their respective sensitivity and specificity values, and positive and negative predictive values are shown in Supplementary Table 3. The BAI remained positively associated with a very high and high risk of CHD following adjustment for potential confounding factors in all the strata and multivariable models (p < 0.05), with the exception of model 3 for a very high risk of CHD in younger women (p ¼ 0.06). In this group, although the magnitude of the association was high (PR ¼ 1.73), the frequency of participants with a very high risk of CHD was low (0.8%), resulting in greater confidence intervals (95%CI). When analyzing high risk, this did not occur, i.e. the association with BAI was maintained even following additional adjustment for WHR. Even considering the possibility of a certain degree of collinearity between WHR and BAI, despite a loss of magnitude in the principal association, a risk of coronary disease was more prevalent in those participants with a higher BAI, irrespective of adjustment for WHR. The magnitude of the associations between BAI and coronary risk was greater in women than in men and greater in the younger age group compared to the older age group for both sexes, both in the case of a 20% or greater risk and of a 10% or greater risk of CHD (Table 3). 4. Discussion The individuals with higher BAI values had a greater 10-year risk of developing CHD compared to those with lower values. The associations remained statistically significant even following adjustment for the potential confounding factors in all the strata. The FRS is probably the global risk score most commonly used to calculate the likelihood of future coronary risk. Nevertheless, few studies have evaluated the association between BAI and global coronary risk in other populations. Bennasar-Veny et al. [31] found a positive correlation between BAI and the FRS (r ¼ 0.125; p < 0.01) in a sample of 50,254 Spanish workers; however, this correlation was weaker than those found for other anthropometric indicators (waist circumference, WHR, waist-to-height ratio and BMI). In that same study, men and women were not analyzed separately, which could have introduced a bias into the findings. Similarly, Beraldo et al. [26] found a positive association between BAI and CVD in a sample of 448 Brazilian HIV patients (OR ¼ 2.76; 95%CI: 1.45e5.27 in the men and OR ¼ 3.44; 95%CI: 1.84e6.45 in the women). Although those authors failed to analyze the association between BAI and FRS, they considered FRS >10% as one of the criteria with which to classify CVD risk. Despite the lack of studies on the association between BAI and global coronary risk, other authors have already demonstrated the association between BAI and cardiometabolic risk factors. In crosssectional studies, the highest BAI value was positively associated with hypertension in adults, even following adjustment for possible confounding factors, in Indian men [14], in adult Brazilians [17] and in elderly Brazilian men and women [15], although the size of the effect was small in that latter group (PR ¼ 1.010; 95%CI: 1.002e1.018; p ¼ 0.016 in the women and PR ¼ 1.020; 95%CI: 1.001e1.040; p ¼ 0.049 in the men). Likewise, an association was found with diabetes in black and white men and women from four North American communities [18], in residents of an urban area of a state capital in southeastern Brazil and in a Brazilian indigenous

Please cite this article as: Almeida RT et al., Association between body adiposity index and coronary risk in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil), Clinical Nutrition, https://doi.org/10.1016/j.clnu.2019.06.001

R.T. Almeida et al. / Clinical Nutrition xxx (xxxx) xxx

5

Table 1 Characteristics of the participants according to sex and age group - ELSA-Brasil, 2008e2010. Men

Women

35e59 years (n ¼ 5,194) c

Age (years) e mean (SD) Body adiposity index e mean (SD)b Very high risk of CHD: 20% [N (%)]a High risk of CHD: 10% [N (%)]a High total cholesterol [N (%)]a Low HDL-cholesterol [N (%)]a High blood pressure [N (%)]a Diabetes Mellitus [N (%)]a Current Smoker [N (%)]a Ethnicity/skin colora Black/brown [N (%)] Education levela Elementary/high school [N (%)] Leisure time physical activitya Insufficiently active [N (%)] Dietary patterna Bakery [N (%)] Low sugar/low fat [N (%)] Fruit and vegetables [N (%)] Traditional [N (%)]

60e74 years (n ¼ 1,487) y

35e59 years (n ¼ 6,304)

60e74 years (n ¼ 1,688)

48.4 (6.4) 26.3 (3.7)* 328 (6.3)* 1,657 (31.9)* 341 (6.6) 181 (3.5)* 266 (5.1)* 998 (19.2)* 780 (15.0)

65.6 (4.3)* 26.9 (3.7)*y 600 (40.4)*y 1,250 (84.1)*y 79 (5.3)* 63 (4.2)* 129 (8.7)y 553 (37.2)*y 116 (11.2)*y

48.6 (6.3) 33.2 (5.5) 49 (0.8) 530 (8.4) 377 (6.0) 38 (0.6) 117 (1.9) 863 (13.7) 831 (13.2)

64.8 (3.9)y 34.7 (5.9)y 102 (6.0)y 596 (35.3)y 169 (10.0)y 14 (0.8) 105 (6.2)y 473 (28.0)y 131 (7.8)y

2,428 (46.8)

501 (33.7)*y

2,889 (45.8)

693 (41.1)y

2,712 (52.2)*

604 (40.6)*y

2,819 (44.7)

824 (48.8)

2,960 (57.0)*

779 (52.4)*y

4,275 (67.8)

1,016 (60.2)y

1,666 (32.1)* 104 (2.0) 823 (15.8) 2,601 (50.1)

395 (26.6)*y 66 (4.4) 466 (31.3) 560 (37.7)

1,241 (19.7) 322 (5.1) 1,708 (27.1) 3,033 (48.1)

293 137 762 496

(17.4)y (8.1) (45.1) (29.4)

a,b,c

Comparative tests for independent samples. p  0.001 compared to women of the same age group. p  0.001 compared to participants of the same sex in the younger age group. a Pearson Chi-squared test for categorical variables. b Student's t-test for continuous variables with normal distribution and unequal variances (except for the comparison of BAI in men with equal variances). c ManneWhitney U test for non-parametric distributions.

* y

Table 2 Cut-off points and validation of the body adiposity index for detection of the risk of coronary heart disease, according to sex and age group. ELSA-Brasil, 2008e2010. Cut-off pointsa

Sens (%)

Spec (%)

AUC (95%CI)

Sens (%)

Spec (%)

AUC (95%CI)

Men 35e59 years Participants (n ¼ 5,354) Very high risk: 20% risk of CHD High risk: 10% risk of CHD

28.0 26.0

43.2 61.9

72.3 54.2

Random Subset (n ¼ 550) 0.602 (0.571e0.633)y 0.611 (0.595e0.627)

55.6 55.0

74.8 52.4

0.649 (0.536e0.763) 0.563 (0.511e0.614)

Men 60e74 years Participants (n ¼ 1,527) Very high risk: 20% risk of CHD High risk: 10% risk of CHD

28.0 28.0

40.8 37.2

69.1 77.3

Random Subset (n ¼ 159) 0.555 (0.526e0.585) 0.587 (0.549e0.625)

34.4 34.1

69.5 77.8

0.596 (0.507e0.686) 0.594 (0.483e0.705)b

Women 35e59 years Participants (n ¼ 6,478) Very high risk: 20% risk of CHD High risk: 10% risk of CHD

34.0 34.0

65.4 55.7

61.7 63.1

Random Subset (n ¼ 663) 0.672 (0.600e0.744) 0.624 (0.598e0.649)

100.0 57.1

0.629 (0.573e0.685)* 0.613 (0.585e0.640)

57.2 50.0

62.5 64.0

0.663 (e - 1.000) 0.618 (0.530e0.706)

Women 60e74 years Participants (n ¼ 1,733) Very high risk: 20% risk of CHD High risk: 10% risk of CHD

36.0 34.0

50.9 59.4

67.1 57.6

Random Subset (n ¼ 169) 69.8 63.9

0.757 (0.561e0.952) 0.616 (0.523e0.709)

95% CI: 95% confidence interval. Sens: sensitivity. Spec: specificity. AUC: area under the ROC curve. CHD: coronary heart disease. Comparison of the areas under the ROC curves: *p  0.05 compared to men of the same age group. yp  0.05 compared to participants of the same sex in the older age group. a Optimal cut-off point according to the Youden index. b 95% CI of the AUC <0.50.

population [20]. BAI was also associated with insulin resistance in male and female patients with acute coronary syndrome [22], with the metabolic syndrome in postmenopausal Filipina and Caucasian women [23] and with two or more risk factors for cardiovascular disease in a North American population sample consisting of black and white individuals of both sexes [2]. On the other hand, following adjustment for other predictive factors, BAI was not found to be associated with the metabolic syndrome in Iranian men or women [25] or in postmenopausal

Brazilian [24] or African-American women [23], or with hyperglycemia in elderly Brazilian men and women [40]. Longitudinally, an association was found between BAI and the incidence of hypertension in North American men and women [16], and with the incidence of diabetes in German men and women [19] and in Indian men [21]. High BAI was also associated with mortality from cardiovascular disease in Australians [28]. Conversely, no association was found with the incidence of diabetes in Indian women [21] or with all-

Please cite this article as: Almeida RT et al., Association between body adiposity index and coronary risk in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil), Clinical Nutrition, https://doi.org/10.1016/j.clnu.2019.06.001

6

R.T. Almeida et al. / Clinical Nutrition xxx (xxxx) xxx

Fig. 1. Areas under the ROC curves for body adiposity index for the identification of coronary risk, according to sex and age group. ELSA-Brasil (2008e2010).

cause mortality or death from cardiovascular disease in North American men [29]. Various factors may interfere with these associations. The findings of the present study identified poorer schooling as an important risk factor in all the strata. In general, the practice of leisure time physical activity and dietary pattern constituted protective factors against the development of CHD over a 10-year

period. Self-reported ethnicity/skin color of black or brown did not represent an important confounding factor. The predictive models identified in epidemiological studies on BAI included age, physical activity, schooling and smoking as potential confounding factors to be controlled [14e16,18e26,28,29,35]. The influence of ethnicity/skin color on the association of BAI with cardiovascular risk factors has

Please cite this article as: Almeida RT et al., Association between body adiposity index and coronary risk in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil), Clinical Nutrition, https://doi.org/10.1016/j.clnu.2019.06.001

R.T. Almeida et al. / Clinical Nutrition xxx (xxxx) xxx

7

Table 3 Association between risk of coronary heart disease and body adiposity index (BAI), adjusted for waist-to-hip ratio and other predictive factors, according to sex and age group. ELSA-Brasil, 2008e2010. Unadjusted PR (95% CI)

Multivariate Model 1 PR (95% CI)

Very high risk of coronary heart disease (≥20%) Men 35e59 years 1.89 (1.53e2.33) (n ¼ 5,194) 1.29 (1.14e1.46) Men 60e74 years (n ¼ 1,487) Women 35e59 years 2.79 (1.56e4.97) (n ¼ 6,304) Women 60e74 years 1.98 (1.36e2.88) (n ¼ 1,688) High risk of coronary heart disease (≥10%) Men 35e59 years 1.56 (1.43e1.69) (n ¼ 5,194) Men 60e74 years 1.10 (1.06e1.15) (n ¼ 1,487) Women 35e59 years 2.00 (1.70e2.36) (n ¼ 6,304) Women 60e74 years 1.56 (1.37e1.78) (n ¼ 1,688)

Model 2 AIC

P

PR (95% CI)

Model 3 AIC

P

PR (95% CI)

AIC

P

1.64 (1.33e2.03)# 2336.826 1.000 1.51 (1.22e1.86)# 2373.577 0.999 1.34 (1.08e1.66)* 2278.120 1.000 1.26 (1.12e1.43)# 2265.541 1.000 1.25 (1.10e1.42)# 2279.948 1.000 1.23 (1.09e1.39)* 2262.670 1.000 2.42 (1.36e4.31)* 554.1372 0.170 1.86 (1.04e3.33)* 512.9969 0.928 1.73 (0.97e3.07)y

506.4127 <0.01

1.77 (1.22e2.58)* 758.4074 0.971 1.70 (1.16e2.49)* 744.2947 0.998 1.57 (1.08e2.31)* 738.4950 1.000

1.47 (1.36e1.60)# 6913.717 1.000 1.25 (1.15e1.37)# 6790.589 1.000 1.20 (1.10e1.30)# 6697.042 1.000 1.10 (1.05e1.15)# 2942.544 1.000 1.08 (1.03e1.12)# 2930.118 1.000 1.07 (1.03e1.12)# 2937.853 1.000 1.82 (1.54e2.15)# 3563.221 1.000 1.48 (1.26e1.75)# 3325.799 1.000 1.41 (1.20e1.66)# 3289.801 1.000 1.42 (1.25e1.62)# 2369.233 1.000 1.38 (1.21e1.57)# 2335.672 1.000 1.29 (1.13e1.47)# 2310.154 1.000

PR: prevalence ratio; 95%CI: 95% confidence interval; AIC: Akaike Information Criterion; P: Pearson residual; *p  0.05; #p  0.001; yp ¼ 0.06. The cut-off thresholds of body adiposity index (BAI) to indicate subjects at coronary risk were as follows: BAI  28.0 for men, except for high risk in younger men (BAI  26.0); and BAI  34.0 for women, except for very high risk in older women (BAI  36.0). Model 1: adjusted for potential confounding factors in each subgroup (education level, leisure time physical activity and dietary pattern for men; education level for younger women; education level and leisure time physical activity for older women). Model 2: adjusted only for waist-to-hip ratio. Model 3: adjusted for variables in model 1 plus waist-to-hip ratio.

seldom been studied [2,10]; however, it is known that the components of body composition can vary as a function of these characteristics [41]. The effect of diet on the association between BAI and cardiometabolic risk factors has not been analyzed in populationbased studies; nevertheless, dietary behavior is known to play an important role in the onset of obesity and, consequently, to contribute to the development of metabolic diseases [42]. Although its measurement represents a challenge to epidemiology, international organizations have encouraged studies on dietary patterns, since analysis of individual nutrients may increase the strength of an association or, conversely, may fail to identify a possible association between overall diet and the risk of chronic diseases [37]. Existing evidence is unable to provide an answer regarding how and in what way measurements of central obesity such as waist circumference and WHR are associated with BMI, a measure of general obesity, in large populations of different ethnicities, and whether there is an interaction between these indicators (or general and central obesity) for the development of coronary disease. The association between BAI and future coronary risk did not depend on WHR in the participants of the ELSA-Brasil. Despite the appearance of new anthropometric indicators and the growing number of studies on their accuracy over recent decades, the identification of a more appropriate universal anthropometric indicator for use in screening for coronary risk in large populations remains a challenge in epidemiology. Nevertheless, anthropometric indicators that include the measurement of hip circumference in their calculation appear to show good potential as predictors of health risks, especially in large population groups. Hip size could reflect the differences in body composition between men and women [1]. In the present study, the optimal cut-off points of BAI for predicting coronary risk were: 28.0 for men of both age-groups, 34.0 in younger women and 36.0 in older women for a 20% or greater risk of CHD; and 26.0 in younger men, 28.0 in older men and 34.0 in

women of both age groups for a 10% or greater risk of CHD. To the best of our knowledge, there are no other studies in which reference values of BAI for the prediction of global coronary risk have been suggested for large populations. Nevertheless, these results are supported by other published evidence in which the recommended cut-off points ranged from 25.15 and 28.75 in men [13,15,17,18,20,21,25,26,40,43e46] to 30.0 and 37.7 in women [13,17,20,25,26,45,46] for the determination of various cardiometabolic risk factors. The different ethnic components, ages and the anthropometric characteristics of each population may explain the differences found. Furthermore, different single endpoints were evaluated (risk factors for T2DM, hypertension and hypertriglyceridemia) rather than an overall risk score such as that evaluated in this study. Another possible explanation is that the authors did not adopt the same criteria with which to define the cut-off points, which are arbitrary and can change depending on what is being studied and as a function of prior knowledge on treatment and preventive methods that already exist [47]. The results of the present analysis should be interpreted with caution. First, since this is a cross-sectional study, it is impossible to affirm whether the participants with the higher BAI values in this cohort will develop coronary artery disease in the future. However, prospective monitoring of this population will result in better evaluation of whether the BAI is indeed a good predictor of coronary risk. Secondly, the ELSA-Brasil sample is not representative of the Brazilian population as a whole, as it does not include the groups at the extreme ends of the spectrum such as the very rich and the very poor, for example. Nevertheless, the sample is large, and since the study is being conducted in several centers, the population is heterogeneous enough to provide new information on the association between BAI and coronary risk in admixed populations. Finally, the effect of other potential confounding factors that could have had an impact on the associations identified was not tested, such as, for example, alcohol abuse and smoking,

Please cite this article as: Almeida RT et al., Association between body adiposity index and coronary risk in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil), Clinical Nutrition, https://doi.org/10.1016/j.clnu.2019.06.001

8

R.T. Almeida et al. / Clinical Nutrition xxx (xxxx) xxx

although the FRS takes reported current cigarette smoking into consideration. In conclusion, higher BAI levels may be associated with a greater risk of developing coronary disease in the future in men and women of both age groups (35e59 years and 60e74 years). BAI proved to be a useful tool for screening for coronary risk, showing a reasonable ability to predict coronary risk in participants of the ELSA-Brasil (AUC > 60%, except for the group of men of 60e74 years of age). The limitations of this indicator are similar to those found for other anthropometric indicators used to screen for adverse health risks. One advantage of the BAI in relation to BMI is the fact that its calculation does not involve the need to measure body weight, thus eliminating the need for scales, which could be useful in field studies or even in clinical practice in precarious settings or whenever the instrument is not available. Furthermore, irrespective of age group, the cut-off points of the BAI for the identification of coronary risk were higher for the women than for the men, suggesting that it would be inappropriate to use the same value for both sexes.

[4]

[5]

[6]

[7]

[8]

[9]

[10]

Funding sources [11]

The ELSA-Brasil baseline study was supported by the Brazilian Ministry of Health (Department of Science and Technology) and the Ministry of Science and Technology (Study and Project Funding agency-FINEP and National Research Council-CNPq) (grants 01 06 0010.00 RS, 01 06 0212.00 BA, 01 06 0300.00 ES, 01 06 0278.00 MG, 01 06 0115.00 SP, and 01 06 0071.00 RJ). The funding source played no role in the study design, data collection, analysis and interpretation, preparation of the manuscript, or in the decision to publish.

[12]

[13]

[14]

Statement of authorship [15]

RTA was the principal author responsible for data analysis and for writing the manuscript. RTA, SMAM and EMLA contributed to the conception and the design of the study. ACP, MJMF, SMAM and EMLA contributed to analysis, interpretation and acquisition of the data. EMLA was responsible for overseeing and leading the planning and execution of this research activity. All the authors contributed to and approved the final version of this manuscript. Conflict of interest All the authors declare that they have no competing interests. Acknowledgements The authors thank the staff and participants of the Elsa Study for their important contributions.

[16]

[17]

[18]

[19]

[20]

[21]

Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.clnu.2019.06.001.

[22]

[23]

References [1] Bergman RN, Stefanovski D, Buchanan TA, Sumner AE, Reynolds JC, Sebring NG, et al. A better index of body adiposity. Obesity 2011;19:1083e9. https://doi.org/10.1038/oby.2011.38. [2] Barreira TV, Staiano AE, Harrington DM, Heymsfield SB, Smith SR, Bouchard C, et al. Anthropometric correlates of total body fat, abdominal adiposity, and cardiovascular disease risk factors in a biracial sample of men and women. Mayo Clin Proc 2012;87:452e60. https://doi.org/10.1016/j.mayocp.2011.12.017. [3] Johnson W, Chumlea WC, Czerwinski SA, Demerath EW. Concordance of the recently published body adiposity index with measured body fat percent in

[24]

[25]

[26]

European-American adults. Obesity 2012;20:900e3. https://doi.org/10.1038/ oby.2011.346. Freedman DS, Thornton JC, Pi-Sunyer FX, Heymsfield SB, Wang J, Pierson RN, et al. The body adiposity index (hip circumference ÷ height (1.5)) is not a more accurate measure of adiposity than is BMI, waist circumference, or hip circumference. Obesity 2012;20:2438e44. https://doi.org/10.1038/ oby.2012.81. Lam BCC, Lim SC, Wong MTK, Shum E, Ho CY, Bosco JIE, et al. A method comparison study to validate a novel parameter of obesity, the body adiposity index, in Chinese subjects. Obesity 2013;21:E634e9. https://doi.org/10.1002/ oby.20504. Lichtash CT, Cui J, Guo X, Chen Y-DI, Hsueh WA, Rotter JI, et al. Body adiposity index versus body mass index and other anthropometric traits as correlates of cardiometabolic risk factors. PLoS One 2013;8:e65954. https://doi.org/ 10.1371/journal.pone.0065954. Kuhn PC, Vieira Filho JPB, Franco L, Dal Fabbro A, Franco LJ, Moises RS. Evaluation of body adiposity index (Bai) to estimate percent body fat in an indigenous population. Clin Nutr 2014;33:287e90. https://doi.org/10.1016/ j.clnu.2013.04.021. Sun G, Cahill F, Gulliver W, Yi Y, Xie Y, Bridger T, et al. Concordance of Bai and BMI with DXA in the newfoundland population. Obesity (Silver Spring) 2013;21:499e503. https://doi.org/10.1002/oby.20009. Fedewa MV, Nickerson BS, Esco MR. Associations of body adiposity index, waist circumference, and body mass index in young adults. Clin Nutr 2018. https://doi.org/10.1016/j.clnu.2018.03.014. Epub ahead. Barreira TV, Harrington DM, Staiano AE, Heymsfield SB, Katzmarzyk PT. Body adiposity index, body mass index, and body fat in white and black adults. J Am Med Assoc 2011;306:828e30. https://doi.org/10.1001/jama.2011.1189.  pez AA, Cespedes ML, Vicente T, Tomas M, Bennasar-Veny M, Tauler P, et al. Lo Body adiposity index utilization in a Spanish Mediterranean population: comparison with the body mass index. PLoS One 2012;7:e35281. https:// doi.org/10.1371/journal.pone.0035281. Freedman DS, Blanck HM, Dietz WH, DasMahapatra P, Srinivasan SR, Berenson GS. Is the body adiposity index (hip circumference/height(1.5)) more strongly related to skinfold thicknesses and risk factor levels than is BMI? The Bogalusa Heart Study. Br J Nutr 2013;109:338e45. https://doi.org/ 10.1017/S0007114512000979. Gupta S, Kapoor S. Body adiposity index: its relevance and validity in assessing body fatness of adults. ISRN Obes 2014;2014:243294. https:// doi.org/10.1155/2014/243294. Chakraborty R, Bose K. Comparison of body adiposity indices in predicting blood pressure and hypertension among slum-dwelling men in Kolkata, India. Malays J Nutr 2012;18:319e28. Leal Neto J de S, Coqueiro R da S, Freitas RS, Fernandes MH, Oliveira DS, Barbosa AR. Anthropometric indicators of obesity as screening tools for high blood pressure in the elderly. Int J Nurs Pract 2013;19:360e7. https://doi.org/ 10.1111/ijn.12085. ~ a-Romero V, Lee DC, Blair SN. Body Moliner-Urdiales D, Artero EG, Sui X, Espan adiposity index and incident hypertension: the aerobics center longitudinal study. Nutr Metab Cardiovasc Dis 2014;24:969e75. https://doi.org/10.1016/ j.numecd.2014.03.004. Oliveira CM, Ulbrich AZ, Neves FS, Dias FAL, Horimoto ARVR, Krieger JE, et al. Association between anthropometric indicators of adiposity and hypertension in a Brazilian population: Baependi Heart Study. PLoS One 2017;12:e0185225. https://doi.org/10.1371/journal.pone.0185225. Hardy DS, Stallings DT, Garvin JT, Gachupin FC, Xu H, Racette SB. Anthropometric discriminators of type 2 diabetes among White and Black American adults. J Diabetes 2017;9:296e307. https://doi.org/10.1111/1753-0407.12416. €ring HU, Schick F, Zierer A, et al. Body Schulze MB, Thorand B, Fritsche A, Ha adiposity index, body fat content and incidence of type 2 diabetes. Diabetologia 2012;55:1660e7. https://doi.org/10.1007/s00125-012-2499-z. Alvim RDOR de O, Mourao-Junior CA, Oliveira CM de, Krieger JE, Mill JG, Pereira AC, et al. Body mass index, waist circumference, body adiposity index, and risk for type 2 diabetes in two populations in Brazil: general and Amerindian. PLoS One 2014;9:e100223. https://doi.org/10.1371/journal.pone.0100223. Talaei M, Sadeghi M, Marshall T, Thomas GN, Iranipour R, Nazarat N, et al. Anthropometric indices predicting incident type 2 diabetes in an Iranian population: the Isfahan Cohort Study. Diabetes Metab 2013;39:424e31. https://doi.org/10.1016/j.diabet.2013.04.001. Marcadenti A, Oliveira VG de, Bertoni VM, Wittke E, Dourado LP, Souza RB de, ^ncia a  Insulina e Indicadores Antropome tricos em Pacientes com et al. Resiste Síndrome Coronariana Aguda. Rev Bras Cardiol 2013;26:259e66. Djibo DA, Araneta MRG, Kritz-Silverstein D, Barrett-Connor E, Wooten W. Body adiposity index as a risk factor for the metabolic syndrome in postmenopausal Caucasian, African American, and Filipina women. Diabetes Metab Syndr 2015;9:108e13. https://doi.org/10.1016/j.dsx.2014.04.011. Gadelha AB, Myers J, Moreira S, Dutra MT, Safons MP, Lima RM. Comparison of adiposity indices and cut-off values in the prediction of metabolic syndrome in postmenopausal women. Diabetes Metab Syndr 2016;10:143e8. https:// doi.org/10.1016/j.dsx.2016.01.005. Motamed N, Rabiee B, Keyvani H, Hemasi GR, Khonsari M, Saeedian FS, et al. The best obesity indices to discriminate type 2 diabetes mellitus. Metab Syndrome Relat Disord 2016;14:249e53. https://doi.org/10.1089/met.2015.0133. Beraldo RA, Meliscki GC, Silva BR, Navarro AM, Bollela VR, Schmidt A, et al. Anthropometric measures of central adiposity are highly concordant with

Please cite this article as: Almeida RT et al., Association between body adiposity index and coronary risk in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil), Clinical Nutrition, https://doi.org/10.1016/j.clnu.2019.06.001

R.T. Almeida et al. / Clinical Nutrition xxx (xxxx) xxx

[27]

[28]

[29]

[30]

[31]

[32]

[33]

[34]

[35]

[36]

predictors of cardiovascular disease risk in HIV patients. Am J Clin Nutr 2018;107:883e93. https://doi.org/10.1093/ajcn/nqy049. Raposo L, Severo M, Santos AC. Adiposity cut-off points for cardiovascular disease and diabetes risk in the Portuguese population: the PORMETS study. PLoS One 2018;13:e0191641. https://doi.org/10.1371/journal.pone.0191641. Dhaliwal SS, Welborn TA, Goh LGH, Howat PA. Obesity as assessed by body adiposity index and multivariable cardiovascular disease risk. PLoS One 2014;9:e94560. https://doi.org/10.1371/journal.pone.0094560. ~ a-Romero V, Sui X, Blair SN. Moliner-Urdiales D, Artero EG, Lee D-C, Espan Body adiposity index and all-cause and cardiovascular disease mortality in men. Obesity 2013;21:1870e6. https://doi.org/10.1002/oby.20399. Wang F, Chen Y, Chang Y, Sun G, Sun Y. New anthropometric indices or old ones: which perform better in estimating cardiovascular risks in Chinese adults. BMC Cardiovasc Disord 2018;18. https://doi.org/10.1186/s12872-0180754-z. Bennasar-Veny M, Lopez-Gonzalez AA, Tauler P, Cespedes ML, Vicente~ ez A, et al. Body adiposity index and cardiovascular health risk Herrero T, Yan factors in caucasians: a comparison with the body mass index and others. PLoS One 2013;8:e63999. https://doi.org/10.1371/journal.pone.0063999. Heitmann BL, Lissner L. Hip Hip Hurrah! Hip size inversely related to heart disease and total mortality. Obes Rev 2011;12:478e81. https://doi.org/ 10.1111/j.1467-789X.2010.00794.x. Vazquez G, Duval S, Jacobs DR, Silventoinen K. Comparison of body mass index, waist circumference, and waist/hip ratio in predicting incident diabetes: a meta-analysis. Epidemiol Rev 2007;29:115e28. https://doi.org/10.1093/ epirev/mxm008. Barzi F, Woodward M, Czernichow S, Lee CMY, Kang JH, Janus E, et al. The discrimination of dyslipidaemia using anthropometric measures in ethnically diverse populations of the Asia-Pacific Region: the Obesity in Asia Collaboration. Obes Rev 2010;11:127e36. https://doi.org/10.1111/j.1467-789X.2009.00605.x. Coqueiro R da S, Santos GAF, Borges LJ, Ferreira T de S, Fernandes MH, Barbosa AR. Anthropometric indicators of obesity and hyperglycaemia in Brazilian older people. J Diabetes Nurs 2013;17:351e5. Aquino EML, Barreto SM, Bensenor IM, Carvalho MS, Chor D, Duncan BB, et al. Brazilian longitudinal study of adult health (ELSA-Brasil): objectives and design. Am J Epidemiol 2012;175:315e24. https://doi.org/10.1093/aje/ kwr294.

9

[37] Cardoso L de O, Carvalho MS, Cruz OG, Melere C, Luft VC, Molina M del CB, et al. Eating patterns in the Brazilian longitudinal study of adult health (ELSABrasil): an exploratory analysis. Cad Saude Publica 2016;32:e00066215. https://doi.org/10.1590/0102-311X00066215. [38] World Health Organization. Waist circumference and waistehip ratio: report of a WHO expert consultation. Geneva. 2008. [39] Wilson PW, D'Agostino RB, Levy D, Belanger AM, Silbershatz H, Kannel WB. Prediction of coronary heart disease using risk factor categories. Circulation 1998;97:1837e47. https://doi.org/10.1161/01.cir.97.18.1837. [40] Alves Júnior CAS, Coqueiro R da S, Carneiro JAO, Pereira R, Barbosa AR, ~es ACM de, et al. Anthropometric indicators in hypertriglyceridemia Magalha discrimination: application as screening tools in older adults. J Nurs Meas 2016;24:215e25. https://doi.org/10.1891/1061-3749.24.2.215. [41] Heyward V. ASEP methods recommendation: body composition assessment. J Exerc Physiol Online 2001;4:1e12. [42] Smith TW, Orleans CT, Jenkins CD. Prevention and health promotion: decades of progress, new challenges, and an emerging agenda. Health Psychol 2004;23:126e31. https://doi.org/10.1037/0278-6133.23.2.126. [43] Chen B-D, He C-H, Ma Y-T, Yang Y-N, Liu F, Pan S, et al. Best anthropometric and atherogenic predictors of metabolic syndrome in the Chinese Han population in Xinjiang: the cardiovascular risk survey. Ann Nutr Metab 2014;65: 280e8. https://doi.org/10.1159/000366427. [44] Xiao X, Liu YY, Sun C, Gang X, Cheng J, Tian S, et al. Evaluation of different obesity indices as predictors of type 2 diabetes mellitus in a Chinese population. J Diabetes 2015;7:386e92. https://doi.org/10.1111/1753-0407.12201. [45] Chen B-D, Yang Y-N, Ma Y-T, Pan S, He C-H, Liu F, et al. Waist-to-height ratio and triglycerides/high-density lipoprotein cholesterol were the optimal predictors of metabolic syndrome in Uighur men and women in Xinjiang, China. Metab Syndrome Relat Disord 2015;13:214e20. https://doi.org/10.1089/ met.2014.0146. [46] Zhang X-H, Zhang M, He J, Yan Y-Z, Ma J-L, Wang K, et al. Comparison of anthropometric and atherogenic indices as screening tools of metabolic syndrome in the Kazakh adult population in Xinjiang. Int J Environ Res Publ Health 2016;13:428. https://doi.org/10.3390/ijerph13040428. [47] Almeida RT de, Almeida MMG de, Araújo TM. Abdominal obesity and cardiovascular risk: performance of anthropometric indexes in women. Arq Bras Cardiol 2009;92:375e80. https://doi.org/10.1590/S0066-782X2009000500007.

Please cite this article as: Almeida RT et al., Association between body adiposity index and coronary risk in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil), Clinical Nutrition, https://doi.org/10.1016/j.clnu.2019.06.001